Human Activity Recognition Based On Inertial Sensors

In recent years,as the improvement of wearable computing,inertial sensor based human activity recognition has become an active research topic in the ubiquitous computing field. In comparison with the conventional vision based method, this one can provide better real time performance, lower limitation and wider application range. The process of Inertial sensor based human activity recognition is collecting data of human motion using multiple inertial sensors placed in different parts of the wearer, then data preprocessing, feature extraction and action classification are carried out.Human activity recognition has been concerned by researchers in recent years, but because of the complexity of human motion and the diversity of objective environment, this research still has a lot of very difficult problems to be solved, such as: problems of solving public activity dataset,effective feature extraction method and the design of high efficient classification algorithm. In order to solve the above problems, the research has done in this dissertation is listed as follows:1. This paper analyses human motion, and summarize it's characteristics,on this basis,we classified and defined the basic activities which constitute the complex actions and motions.2. Characterize the human activity by using human joint angle have been proposed, then used motion capture system based on inertial sensors to collect human action data which are all stored in a new database named MongoDB.3. According to the characteristics of human activity data, we adopt the appropriate methods to preproced the data and extracted their characteristics from the time domain and frequency domain, after that the principal component analysis is used to reduce the dimensions of characteristics vectors.4. Used the hidden markov model,which have gotten successful application in the field of speech recognition to classify the reduced features. Experimental results show that the height estimation algorithm is effective and feasible.